An updated analysis from Gallup this week has revealed another factor that could be behind Republican presidential nominee Donald Trump's popularity: expensive mortgage-interest payments.

According to the analysis, respondents in hundreds of surveys were more likely to view Trump favorably if they lived in Zip codes with heavy mortgage-interest burdens relative to local incomes, after taking into account a range of socioeconomic factors.

“The increasing unaffordability of housing is a major macroeconomic problem in the United States,” said Gallup economist Jonathan Rothwell. “The sort of long-term financial stress wrought through housing, and the way it plays out in people’s lives, could translate into political frustration with the status quo.”

The respondents also were more likely to view Trump favorably if they lived in Zip codes where more of their neighbors received the Earned Income Tax Credit, which the government pays to low-wage workers, or where a greater share of income came in the form of checks from Social Security for the elderly or disabled.

By contrast, those who lived in Zip codes where more income came from stocks and other investments — markers of affluence — were less favorably disposed toward the New York businessman.

Working with his colleague Pablo Diego-Rosell, Rothwell drew on Gallup's interviews with some 125,000 Americans since the summer of 2015 as part of the organization's daily polling on the election, and compared respondents demographically similar in age, gender, race, education, religion, occupation and a range of other factors.

He combined the results of the polls from the Internal Revenue Service on the size of mortgage-interest payments by Zip code, along with the amount of income from investments, Social Security and the tax credit for earned income. Since taxpayers can save money by claiming a deduction for mortgage interest, the federal data gave the two researchers a good idea of the size of a neighborhood's mortgages relative to residents' incomes.

This burden did not appear to be a result of the housing crisis in 2008. The economists did not find evidence of a relationship between support for Trump and increases or decreases in local housing prices. They also could not find a relationship between Trump's popularity and consumer debt, including car loans and credit-card balances.

Rather, Rothwell said the burden of residential debt in the communities where Trump is winning more likely reflects broader economic trends. Even before the financial crisis, the price of housing had been increasing for years while typical household incomes were not.

A mortgage is not always evidence of financial distress. Often, a household with a costly mortgage is better off than a household without one. A mortgage indicates that a family is not paying monthly rent to keep a roof over their heads. For many Americans, owning a home represents membership in the middle class.

In the communities where Trump is viewed favorably, however, many homes might be sources of worry for their owners rather than comfort.

The data from Gallup contribute to a fierce and enduring debate about how much the economy has to do with Donald Trump's appeal.

The bulk of the Republican nominee's supporters are on the margins of the contemporary labor force. His coalition includes many blue-collar men without college degrees. On the other hand, some pundits have argued that economic circumstances are not the reason for Trump's unexpected success in the campaign so far, and that his voters are drawn by his appeals to racial anxiety.

Rothwell's research shows that while those who view Trump favorably are much more worried about money than those who do not, this anxiety is difficult to explain based on his supporters' actual personal finances. In most respects, those who are favorably disposed to Trump are at least as well off as the rest of the country.

Even if Trump's supporters have not themselves fallen on hard times, they often live in places where economic opportunity is scarce. Rothwell first uncovered this apparently paradoxical pattern in a previous draft of the working paper. He and Diego-Rosell shared a revision containing the additional findings on mortgages and taxes with The Washington Post this week.

The exact relationship between Trump's supporters and the communities they inhabit remains unclear. Rothwell cautioned that the data were difficult to interpret without more detailed information on the personal finances of Trump's supporters, and that further research could lead to different conclusions.

Whatever the explanation, Rothwell's previous analysis rebutted a widespread theory that Trump's supporters live in areas where globalization's costs have exceeded its benefits. Rothwell found that among voters who were demographically similar, those who lived in areas where the economy was negatively affected by Chinese imports were no more likely to view Trump favorably.

In the revised draft, Rothwell and Diego-Rosell also show that those who work in manufacturing or occupations with a greater share of immigrant workers are not more likely to support Trump, either.

It is true that, on the whole, Trump does modestly better with manufacturing workers than with those in other sectors. Yet these workers share many other traits with the rest of Trump's supporters — for instance, many are men without college degrees. Because those who work in manufacturing do not favor Trump more after taking into account these other similarities, it seems that economic competition from goods manufactured abroad does not account for Trump's success with this group.

Even if globalization is not the reason, other evidence shows that Trump is successful in places afflicted by some kind of social and economic malaise. This pattern is particularly clear in medical data. Trump is more popular where residents are less healthy.

The new findings from Rothwell and Diego-Rosell confirm this point.

Respondents who live in communities with greater mortality view Trump more favorably, Rothwell and Diego-Rosell found. That is also true of respondents who live in neighborhoods with elevated rates of diabetes and disability.

Those who live in places where more babies are born below normal weight, which can be related to ill parental health, are more favorably disposed to Trump.

Rothwell and Diego-Rosell did not find a connection between Trump's popularity and drug overdoses or binge drinking.